Modeling indirect interaction in open computational systems

Abstract

Open systems are part of a paradigm shift from algorithmic to interactive computation. Multiagent systems in nature that exhibit emergent behavior and stigmergy offer inspiration for research in open systems and enabling technologies for collaboration. This contribution distinguishes two types of interaction, directly via messages, and indirectly via persistent observable state changes. Models of collaboration are incomplete if they fail to explicitly represent indirect interaction; a richer set of system behaviors is possible when computational entities interact indirectly, including via analog media, such as the real world, than when interaction is exclusively direct. Indirect interaction is therefore a precondition for certain emergent behaviors.